DATASETHugging Face

5/5

Belebele — Parallel Reading Comprehension Benchmark (Arabic variants)

Arabic(ara)TextQAWest Africa

View Source

Dataset Info

Language

Arabic (ara)

Modality

Text

Domain

General

Size

900

Institution

Lucas Bandarkar; Davis Liang; Benjamin Muller; Mikel Artetxe; Satya Narayan Shukla et al. (Meta AI / UCLA)

Last Updated

20 May 2026

License & Access

SPDX identifierCC-BY-SA-4.0

Open — attribution required; commercial use allowed.

RedistributionYes

PII RISK

Unknown risk

Full Record

32 of 46 fields stated

Every column the registry tracks. Values marked derivedwere filled by a rule or read from the source's text rather than stated in its metadata; the rule or the quoted sentence is shown under the value.

Registry columns (23/27)

iso code: ara
Dataset name: Belebele — Parallel Reading Comprehension Benchmark (Arabic variants)
Modality: Text
Type: dataset
Link: https://huggingface.co/datasets/facebook/belebele
NLP Task: QA
Domain: General
Data Format: json
Size Metric: Questions
Size Value: 900
License: CC-BY-SA-4.0
Open Access: Open
Source Repository: Hugging Face
Year Published: 2023
Creator/Author(s): Lucas Bandarkar; Davis Liang; Benjamin Muller; Mikel Artetxe; Satya Narayan Shukla et al. (Meta AI / UCLA)
Funder/Partner: Meta
Dialect Coverage: MSA/Dialectal (6 Arabic variants)
Gender Balance: Not stated
Speaker Count: Not stated
Annotation Type: Human
Status: Active
Has Datasheet: Yes
PII Risk: Not stated
Consent Verified: Not stated
Verified By: Agboola Blessing
Date Indexed: 2026-05-20
Notes / Flags: 900 QA items per language variant. Arabic variants: arb_Arab (MSA), acm_Arab, arz_Arab, ary_Arab, apc_Arab, arb_Latn. Multiple-choice reading comprehension base…

Quality (DQI) (8/12)

Datasheet Fields Completed: Not stated
SPDX License ID: CC-BY-SA-4.0
Redistribution Allowed?: Yes
Content Hash: 7899cdfa4e1e0d733fd77c848e2c273cb1d32be2derivedHugging Face card: commit sha
Submitted By: hf_data pipeline (automated)
Fetch Date: 2026-05-20
Last Updated Date: 2024-08-12derivedHugging Face card: lastModified
Annotation Guidelines Available?: Not stated
IAA Score: Not stated
Cleaning Notes: Not stated
Benchmark Exists?: Yes
Benchmark Citation: https://huggingface.co/atishay23/whisper-small-ar

Representativeness (DRI) (1/7)

Gender Distribution: Not stated
Region Coverage: Not stated
Region Distribution: Not stated
Age Range Coverage: Not stated
Register: Not stated
Marginalized Group Note: Not stated
Demographic Fields Reported Count: 0

Readiness Score Breakdown

5/5

Data Coverage100%

Model Availability100%

Benchmarks100%

Documentation72%

Recency100%

Representativeness

Gender Balance

Regional Distribution

Dialect Coverage

Inclusivity

Provenance

Source URL

https://huggingface.co/datasets/facebook/belebele

Fetch Timestamp

20 May 2026

Content Hash

7899cdfa4e1e0d733fd77c848e2c273cb1d32be2

Audit Log

Record indexed from Hugging Face

20 May 2026

Source status: Active

20 May 2026

Other Copies

The same resource, also published at these locations. It is counted once.

dataset

belebele

Data format: json

24 languages3 copiesArabic(ara)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesArabic(ara)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

Also Available In

This is the same multilingual corpus, sliced for other languages the platform indexes it under.

dataset

belebele

Data format: json

24 languages3 copiesAfrikaans(afr)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesAmharic(amh)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesBambara(bam)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

Belebele - Nigerian Fulfulde

Multiple-choice reading comprehension benchmark; fuv_Latn configuration confirmed; exact licence is CC-BY-SA-4.0, which is not available in dropdown.

24 languages3 copiesNigerian Fulfulde(fuv)TextQA

HUHugging Face

Open3 Jun 2026

dataset

Belebele - West Central Oromo

Confirmed gaz_Latn configuration; multiple-choice reading comprehension benchmark based on FLORES passages.

24 languages3 copiesWest Central Oromo(gaz)TextQA

HUHugging Face

Open3 Jun 2026

dataset

belebele

Data format: json

24 languages3 copiesHausa(hau)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesIgbo(ibo)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

Belebele - Kikuyu

QA · General domain · Meta AI / Belebele Authors

24 languages3 copiesGikuyu(kik)TextQA

HUHugging Face

Open—

dataset

belebele

Data format: json

24 languages3 copiesKinyarwanda(kin)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

Belebele - Kimbundu

Belebele is a multiple-choice reading comprehension dataset spanning 122 language variants; verify kmb_Latn split in viewer before final registry submission.

24 languages3 copiesKimbundu(kmb)TextQA

HUHugging Face

Open3 Jun 2026

dataset

belebele

Data format: json

24 languages3 copiesLingala(lin)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesLuganda(lug)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesMalagasy(mlg)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

Belebele (Chichewa)

Confirmed via web search: Belebele's 122-language-variant reading-comprehension benchmark includes Nyanja/Chichewa. Not tagged with language:nya in the Huggi…

24 languages3 copiesChewa(nya)TextQA

HUHugging Face

Open3 Aug 2026

dataset

belebele

Data format: json

24 languages3 copiesShona(sna)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesSwahili(swa)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesTigrinya(tir)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesTswana(tsn)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesTsonga(tso)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

dataset

belebele

Data format: json

24 languages3 copiesWolof(wol)Textquestion-answering, zero-shot-classification, text-classification, multiple-choice

HUHugging Face

Open27 Oct 2026

Related Datasets

dataset

Mozilla Common Voice 24.0 — Arabic (ar) Speech Dataset

Common Voice v24.0 Arabic: 136,040 clips, 157.28 hours total, 91.74 hours validated. 1,651 speakers. Text corpus: 1,801,544 sentences. CC0 licence. Includes …

9 languages2 copiesArabic(ara)SpeechASR

OTOther (Mozilla)

Open20 May 2026

dataset

Wojood — Arabic Named Entity Recognition Corpus (WojoodFine)

550K tokens, 75K named entity mentions. Nested and fine-grained NER with 21 entity types and 47+ subtypes (WojoodFine). Also includes WojoodGaza (60K tokens,…

Arabic Speech Corpus (ASC) — Nawar Halabi / University of Southampton

First open-source Arabic speech corpus for TTS synthesis. Recorded in studio (Damascene/South Levantine accent). 1,813 WAV files + TextGrid phoneme labels. A…

4 copiesArabic(ara)SpeechTTS/ASR

HUHugging Face

Open20 May 2026